A large number of hereditary breast cancers are due to mutations in the two breast cancer susceptibility genes, BRCA1 and BRCA2. However, many hereditary breast cancer cases cannot be accounted for by mutations in these genes, and these are believed to be due to as yet unidentified breast cancer predisposition genes (BRCAx). Using complementary DNA microarrays to classify hereditary breast cancers, we show that BRCA1 and BRCA2 tumors display distinct gene expression profiles and can be classified on the basis of gene expression patterns. We are analyzing several hereditary breast cancers that have tested negative for mutations in BRCA1 and BRCA2. We tested primary tumors against a microarray of 6,512 genes and employed mathematical analyses to determine a set of genes capable of discriminating mutation-positive breast cancers. Using permutation analysis of multivariate classification functions, we established that the expression profiles of BRCA1 and BRCA2 tumors were significantly different statistically. With analysis of variance between gene expression levels and genotype we ascertained those genes capable of separating BRCA1 from BRCA2 breast tumors. Examination of this gene list suggested functional differences between BRCA1 and BRCA2 tumors, including differences in p53 regulation and homeostasis, and (for BRCA1) the possible establishment of a constitutive stress–type state. We extended our gene expression results to encompass pathological findings by generating a tissue microarray of approximately 200 breast tumor specimens. We are currently determining if BRCAx breast tumors can be classified and subjected to linkage analysis to recognize new breast cancer predisposition loci. Gene expression analysis has for the first time revealed a consistent and significant gene expression difference between BRCA1 and BRCA2 cancers, a finding with therapeutic implications.